Please note, with Q3 QRC (August 2020), SAP Analytics Cloud Smart Predict and SAC Planning will integrate even more closely. With that release, SAP Analytics Cloud Smart Predict will be able to use planning models as data sources and write back generated forecasts directly into those SAC planning models. Planners can easily leverage predictive forecasts at scale, supporting them in making data-driven business decisions. Due to the openness of Smart Predict, planners can understand and trust the underlying time series models.
The SAP Analytics Cloud’s augmented analytics feature that will help with predictive forecasting is called Smart Predict. Smart Predict makes your life easier by predicting events, values, and trends. You can use this feature in a very simple way within stories, or set up predictive scenarios depending on your forecasting needs.
To set up a predictive scenario, head to the create menu and select Predictive Scenario.
There are 3 predictive scenario types you will need to choose from.
Choose this type if you need to make a business decision for a binary event. For example, this could be used to generate predictions for whether individual customers would be likely to buy a new product.
Choose Time Series to forecast numerical values over a set time period. For example, forecasting the sales quantity of shoes by month or week, using historical data.
Choose regression to predict numerical values and explore key values behind them. For example, predicting the price of an imported item based on projected duties or shipping charges.
Create Your Predictive Model
Once you’ve chosen your scenario type, you’ll need to select a data set or planning model that holds the training data for the predictive model. To do so, simply select your choice.
Depending on the scenario type, you will need to select variable roles for columns within your data. Training uses machine learning to build a predictive model based on the information in the data set. For an explanation on how to train a classification or regression predictive model check out this article. To learn how to train a time series model, check out this article.
Debrief your Predictive Model
Once you’ve created your predictive model, it’s time to explore your predictions and understand in detail what the drivers and trends are that led to these findings. SAP Analytics Cloud Smart Predict offers rich tooling to dive into all necessary details of the underlying predictive model. These tools are optimized to the respective model types and differ slightly between predictive scenarios for classification, regression and time series.
Analyze your debrief results to find out the confidence of your prediction, which influencers have the highest influence on your target variable, perform profit simulations, detect any errors, and more. For specific instructions on how to debrief your predictive model check out the following articles:
Prefer to learn by video? Check out the Smart Predict learning playlist. This playlist will guide you step by step through each type of generating a predictive scenario, including how to debrief each type of scenario.
Next you can Apply your predictive output to a dataset, planning model or publish to a Predictive Analytics Integrator (Pai).
Follow the steps in this resource, to learn how to operationalize your predictive model. With the integration of Smart Predict into SAP Analytics Cloud planning in Q3 2020, new additional guides and materials will be made available.
Adding Time-Series Forecasts to Stories
Any time-series chart in a story allows you to predict future outcomes of that same time series. This is as simple as selecting a time-series chart in your story and clicking Add Forecast. You then have the option to select Automatic Forecast, or choose from advanced options of Linear Regression, or Triple Exponential Smoothing.
Your predicted forecast will show up within the chart.
To check forecast quality, simply click the green Forecast button under the title of the chart. You can also change the number of forecast periods displayed from this screen.
To learn more about in story predictive forecasts check out this video.